The question of which side of a SIP call will determine the media selection depends on the directionality of the call and whether early offer or delayed offer is being used. With early offer, the initial INVITE message will carry the SDP offer and the OK message sent by the recipient of the call will have the SDP response with media selection. With delayed offer, the OK message by the recipient of the call holds the SDP offer, and the ACK by the initiator holds the response with media selection. This is by default, anyway. This process gets more complicated with re-invites and multiple hops.
If your dump shows the full SIP message exchange between one device and its two neighbors, that is a good place to start. You may need additional traces from additional devices to get a full picture of what is going on. Start at the initiation of the call and work your way step by step.
I'm not familiar with Pexip: Is that the video bridge itself with a front end, or is that the front end and the video bridge is elsewhere? (Because that would be another hop.)
